Custom Software Engineer (Tax Technology Leader)

Answer:

I focus on decoupling and clear boundaries.

• Domain-Driven Design: Break down the monolith by business domains (e.g., Tax Calculations, User Management).
• Go Concurrency: Utilize Go's lightweight goroutines to handle high-throughput requests concurrently.
• API Gateway: Use an API gateway to route requests, handle authentication (JWT), and rate-limit.
• Containerization: Dockerize each Go service and orchestrate with Kubernetes. Use Helm charts to manage deployments.
• Observability: Integrate structured logging and distributed tracing (like OpenTelemetry) to track requests across services.

Answer:

• Decoupling Services: Instead of synchronous HTTP calls that can cause cascading failures, I use a message broker. For instance, when a tax document is uploaded, a message is published to an SNS topic.
• Asynchronous Processing: Worker services subscribe to SQS queues, pick up the messages, and process the documents asynchronously.
• Resilience: If a worker fails, the message remains in the queue (or goes to a Dead Letter Queue after retries) ensuring no data loss.
• Idempotency: I design consumers to be idempotent so processing the same message twice doesn't cause side effects.

Answer:

• Terraform: I use Terraform to provision cloud resources declaratively (VPCs, RDS instances, EKS clusters). It keeps infrastructure version-controlled and reproducible.
• State Management: Store Terraform state remotely (e.g., in S3 with DynamoDB locking) for team collaboration.
• Helm: For Kubernetes, I write Helm charts to template application deployments. This makes it easy to deploy the same Go service across different environments (dev, staging, prod) by simply passing different values.yaml.

Answer:

(Use a STAR story from your 1Finance experience)

• Situation: We had legacy encryption scripts running slowly or monolithic services becoming hard to maintain.
• Task: I needed to migrate endpoints to a new GoFiber platform and improve response times.
• Action: I designed the new Fiber router groups, implemented DB-backed JWT middleware with Redis caching, and wrote idempotent encryption backfill scripts. I optimized PostgreSQL queries by adding proper indexes.
• Result: We saw a significant drop in latency and successfully decoupled the services without downtime.

Answer:

• Go API Tests: I use table-driven tests for unit testing business logic. For integration tests, I use tools like testcontainers-go to spin up a real PostgreSQL container, run migrations, and test the repository layer.
• UI Tests: For the frontend, I prefer Cypress. I write End-to-End (E2E) tests that simulate user journeys (e.g., logging in, submitting a form). I mock API responses for isolated frontend testing, and run full E2E tests against a staging environment in the CI/CD pipeline.

Answer:

• Authentication/Authorization: Use robust Identity Providers and JWT-based auth. Implement Role-Based Access Control (RBAC) middleware in the Go APIs.
• Data Security: Encrypt data at rest (using AWS KMS or similar) and in transit (TLS). Mask sensitive logs.
• Vulnerability Scanning: Integrate tools in the CI/CD pipeline (e.g., Trivy for Docker images, GoSec for static code analysis) to catch vulnerabilities before deployment.
• Least Privilege: Ensure IAM roles and Kubernetes service accounts have only the permissions they strictly need.